87a3dac115e68d59f4af9cac63a8cbc7

This model is a fine-tuned version of facebook/opt-6.7b on the ccdv/patent-classification [abstract] dataset. It achieves the following results on the evaluation set:

• Loss: 2.2767
• Data Size: 1.0
• Epoch Runtime: 505.2384
• Accuracy: 0.5579
• F1 Macro: 0.5211

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 5e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• distributed_type: multi-GPU
• num_devices: 4
• total_train_batch_size: 32
• total_eval_batch_size: 32
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: constant
• num_epochs: 50

Training results

Training Loss	Epoch	Step	Validation Loss	Data Size	Epoch Runtime	Accuracy	F1 Macro
No log	0	0	2.9897	0	24.1700	0.1006	0.0667
No log	1	781	2.2566	0.0078	28.9549	0.2226	0.0431
No log	2	1562	2.1080	0.0156	43.3543	0.1332	0.0392
No log	3	2343	2.0448	0.0312	64.7618	0.2083	0.0411
0.0536	4	3124	1.8338	0.0625	88.0692	0.2935	0.1572
1.6974	5	3905	1.6295	0.125	131.2229	0.3584	0.2070
1.5444	6	4686	1.5326	0.25	161.0077	0.4197	0.2648
1.3205	7	5467	1.2482	0.5	270.8500	0.5669	0.4136
1.1201	8.0	6248	1.3244	1.0	516.7468	0.5613	0.3907
0.976	9.0	7029	1.2025	1.0	510.4597	0.6024	0.4622
0.7706	10.0	7810	1.3426	1.0	505.6420	0.5863	0.5102
0.6253	11.0	8591	1.4561	1.0	510.1305	0.5765	0.5224
0.4496	12.0	9372	1.8850	1.0	504.6520	0.5705	0.5074
0.2639	13.0	10153	2.2767	1.0	505.2384	0.5579	0.5211

Framework versions

• Transformers 4.57.0
• Pytorch 2.8.0+cu128
• Datasets 4.3.0
• Tokenizers 0.22.1